The analysis and modeling of the response of parts of the body
provides valuable insight into many features of the HRTF. In spatial sound
simulations, partial models, such as the spherical head model, can
also generate simple and effective approximate localization cues.
In this paper, we consider the composition of an approximate HRTF
from the responses of structural components by making use of detailed
measurements of isolated pinnae and of a pinna-less head and torso.
We determine that such a composition is sensitive to additional
geometric parameters that can be obtained from anthropometry. We show
that, with such parameters, simple composition rules lead to a very
good correspondence of measured and composite HRTFs.